Vol. 55, No. 2INFECTION AND IMMUNITY, Feb. 1987, p. 320-3280019-9567/87/020320-09$02.00/0Copyright © 1987, American Society for Microbiology

Selective Pressures and Lipopolysaccharide Subunits asDeterminants of Resistance of Clinical Isolates of Gram-Negative

Bacilli to Human SerumREUVEN PORAT,t MARGARET A. JOHNS, AND WILLIAM R. McCABE*

Maxwell Finland Laboratory for Infectious Diseases, Boston City Hospital, and Department of Medicine, BostonUniversity School of Medicine, Boston, Massachusetts 02118

Received 22 July 1986/Accepted 28 October 1986

Differences in molecular composition of lipopolysaccharides (LPS) between serum-sensitive (S) clinicalisolates ofEscherichia coli and serum-resistant (R) clones derived by serial passage in serum were demonstratedto determine sensitivity or resistance to killing by normal human serum (NHS). LPS from R clones had agreater proportion of higher-molecular-weight, more highly 0-antigen-substituted subunits than LPS fromtheir serum S parents. Utilization of a liposomal model with inserted LPS simulating bacterial cell wallsestablished LPS as the site of serum bactericidal action. Liposomes containing S LPS were lysed, whileliposomes containing R LPS were unaffected by NHS. R and S LPS were fractionated into higher (Fl)- andlower (F2)-molecular-weight fractions. Liposomes containing R LPS or the Fl fraction of S and R LPS werenot lysed by serum. Liposomes containing the F2 fraction of S or R LPS were lysed by serum analogous to thatobserved with liposomes containing intact S LPS. These findings establish LPS to be one site of serumbactericidal activity and demonstrate that the higher-molecular-weight, highly 0-antigen-substituted LPSsubunits mediate resistance to killing by NHS.

Gram-negative bacilli (GNB) have become the most fre-quent cause of nosocomial infections and death from infec-tions in this country during the past 4 decades (11). Complexinterrelations between host defense mechanisms, therapeu-tic measures, and bacterial factors influence the develop-ment and outcome of such infections (1, 11). Althoughadherence mechanisms and exotoxin production appear cru-cial for colonization and infection at some body sites (2, 7,24), other possible virulence factors of GNB are less welldefined. Resistance to rapid killing by normal human serum(NHS) appears important for development of systemic GNBinfections since most isolates from the gastrointestinal tractare sensitive (S), while isolates from systemic clinical infec-tions are almost uniformly resistant (R), 2 85%, to killing byserum (12, 19).Mechanisms responsible for the almost uniform resistance

of GNB isolated from clinical infections and the exactmechanism(s) responsible for killing by NHS have not beendelineated, although several explanations have been pro-posed. Cell wall components, particularly lipopolysaccha-ride (LPS), K or capsular antigens, and outer membraneproteins have been postulated to be sites sensitive to andresponsible for resistance to killing by NHS (14, 16, 17, 22,26, 28). Serum R GNB have been shown to contain morelonger-chain, highly 0-antigen-substituted LPS subunitsthan S strains (4, 5, 22, 25, 26, 28), but laboratory-maintainedor -derived rough mutants rather than fresh clinical isolateshave often been used to represent serum S strains. Goldmanet al. (5) also demonstrated that serum R clones could beselected from laboratory-maintained S strains by serial pas-sage in serum. Others have suggested that plasmid-mediated

* Corresponding author.t Present address: Department of Internal Medicine "D," Ichilov

Hospital, Tel Aviv Medical Center, Tel Aviv, Israel.

alterations in outer membrane proteins are associated withslight increases in serum resistance (14, 16). Capsular anti-gen has also been proposed to mediate resistance (17).Classically, killing of intact bacteria has been the standardassay of the activity of NHS, but the multiple cellularcomponents present hamper exact identification of the pre-cise site of killing.These studies examine the selection of R clones by serial

passage of fresh clinical isolates of S GNB in NHS andutilize the liposomal model of Kinsky et al. (9) to clearlyestablish LPS as a site of serum bactericidal activity. Use ofthis multilamellar liposomal model allowed identification ofLPS subunits responsible for conferring resistance to killing.This model provides an ideal system for examination of therole of various bacterial and serum components involved inserum bactericidal activity.

MATERIALS AND METHODS

Bacterial strains. Clinical isolates of serum S Escherichiacoli (strain, serotype: 39, 04:K-12; 109, 06; 227, 018ac:K1;248, 06:K2; 337, 075; and 348, 07) were obtained from theMicrobiology Laboratory of University Hospital, Boston,Mass. The strains were identified as serum S and wereserotyped as described previously (12).NHS pool. Blood was obtained by venipuncture (50 ml

each) from 30 normal volunteers, allowed to clot, andseparated in the cold, sera were pooled, and 1.0-ml aliquotswere frozen at -60°C. This pool was used in the selection ofserum R clones and as a source of complement and antibodyin serum bactericidal assays.

Selection of R clones. Each serum S E. coli was inoculatedin 2 ml of brain heart infusion (BHI) broth and incubatedovernight at 35°C. Samples (0.001 ml) were inoculated intofresh BHI broth containing 50% NHS and incubated at 35°C

320

on July 12, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

LPS SUBUNITS AS DETERMINANTS OF SERUM RESISTANCE 321

for 2 h. These subcultures were screened for the presence ofserum R clones by the agar plate method of Fierer et al. (3).Bacteria (serum R) growing within the area where NHS hadbeen deposited were inoculated into BHI broth and incu-bated overnight. The same sequence of incubation in 50%NHS, subculturing, and selection by the agar plate methodwas repeated until there was a solid bacterial lawn in the areawhere NHS had been added. Once a serum R population wasobtained, cultures were serially passed 15 to 20 times inserum-free broth to ensure their stability.

Bactericidal assay. Bacterial killing was quantitated by astandard microtiter plate assay technique (12). Bacteriagrown in BHI broth to the log phase (4 h) were dilutedlogarithmically in phosphate-buffered saline containing 5 xlo- M MgCl2 and 10-i M CaCl2 (PBS++). Duplicate 1.0-mlsamples of 10-6 dilutions were plated for counts of theoriginal inoculum. Microtiter plates (Linbro; Flow Labora-tories, Inc., McLean, Va.) were inoculated with 0.05-mlsamples of bacteria (approximately 250), 0.05 ml of PBS++,and 0.1 ml of NHS or decomplemented NHS (56°C for 30min) and incubated for 2 h. Duplicate 0.1-ml samples wereplated overnight for bacterial counts. The percentage of theoriginal inoculum surviving after 2 h of incubation wascalculated.LPS. Each bacterial strain (serum S parent and serum R

clone) was grown in 15 liters of Casamino Acids broth (DifcoLaboratories, Detroit, Mich.) supplemented with yeast ex-tract and glucose (Sigma Chemical Co., St. Louis, Mo.).After 16 to 18 h of incubation, bacteria were killed andprecipitated by the addition of 2.5 volumes of acetone.Precipitated bacteria were washed three times in acetoneand air dried, and LPS was extracted by the hot phenol-water method of Westphal et al. (29), dialyzed, and lyophi-lized.

Radiolabeled LPS. Bacteria were grown with the additionof 3H-labeled sodium acetate (5 mCi/liter; New EnglandNuclear Corp., Boston, Mass.), and LPS was extracted asdescribed above (29).

Saponification of LPS. LPS used for incorporation intoliposomes was saponified by treatment with 0.1 N NaOH (2mg/ml) at 35°C for 18 h. The pH was adjusted to 7.0 to 7.2with 1 N HCI, and the saponified LPS was dialyzed againstdistilled water for 3 days and lyophilized (13).PAGE. Polyacrylamide gel electrophoresis (PAGE) of

LPS was performed in 0.75-mm-thick 14% T gels with Trisglycine buffer (pH 8.4) containing 0.1% sodium dodecylsulfate by the method of Laemmli (10). Samples dissolved ina digestion buffer of 0.18 M Tris (pH 6.8), 90 mM dithiothrei-tol, sodium dodecyl sulfate, and 25% urea were heated to100°C for 10 min or unheated, and 0.01 ml of 1-mg/mlconcentration was applied to gels. Gels were run at aconstant current of 15 mA in an LKB model 2001 verticalelectrophoresis unit, stained with silver reagent, and dried.All chemicals and reagents were of electrophoresis grade(Bio-Rad Laboratories, Richmond, Calif.).Sephadex G-200 chromatography of LPS. LPS was frac-

tionated on a Sephadex G-200 column by a modification ofthe method of Vukajlovich and Morrison (27). LPS fromboth serum S and R bacteria was disaggregated in 0.1 M Trisbuffer (pH 8.5) containing 0.6% sodium deoxycholate (DOC)at 50 mg/ml. A 2-ml sample was applied to a Sephadex G-200column (2.6 by 55 cmn; Pharmacia, Inc., Piscataway, N.J.),equilibrated with the same buffer, and eluted at 35°C at a rateof 12 ml/h. Fractions of 2.5 ml were collected and analyzedfor hexose and 3-deoxy-D-manno-octulosonic acid (KDO)content.

Chemical methods. Hexose was determined by either theanthrone method (23) or the cysteine-sulfuric acid method(30) with glucose as the standard. KDO was determined bythe thiobarbituric acid method (18) with 2-keto-3-deoxyoctonate (Sigma) as a standard.Liposome preparation. Stock solutions of veronal-buffered

saline (VBS), MgCl2-CaCl2, gelatin-VBS (GVBS), andGVBS containing 0.125 ml of MgCl2-CaCl2 (GVBS++) wereprepared. Three lipids (Sigma), dimyristyl phosphatidyl-choline (25 mM in chloroform), cholesterol (75 mM inchloroform), and dicetylphosphate (3 mM in 1:1 [vol/vol]methanol-chloroform), were mixed in a 10-ml pear-shapedboiling flask in a molar ratio of 2:1.5:0.22 as described byKinsky et al. (9) and Kataoka et al. (8). Chloroform andmethanol were removed by evaporation on a rotary evapo-rator (Buchler Instruments Div., Nuclear-Chicago Corp.,Fort Lee, N.J.), producing a thin film of dried lipids on thelower portion of the flask which was placed in a desiccatorjar under vacuum for 1 h to complete drying. Appropriateamounts of LPS and 15 ,uCi of 51Cr (as sodium chromate;New England Nuclear Corp.) in a total volume of 0.2 ml ofGVBS were added, and the flask was agitated vigorously(Vortex Genie mixer; The Vortex Manufacturing Co., Cleve-land, Ohio) for 5 min. The multilamellar liposomes formedwere passed through a column of Sephadex G-25 (1.6 by 30cm; Pharmacia Fine Chemicals, Piscataway, N.J.) inGVBS++ to separate 51Cr-containing liposomes from free5tCr. Fractions (1 ml) were collected and counted in agamma counter (model 1197; Searle Analytic, Des Plaines,Ill.), and the entire 51Cr-labeled liposome fraction waspooled, washed in GVBS, and suspended in 1 ml ofGVBS++. The liposome suspension (200 [lI) was incubatedat 35°C for 30 min with one of the following: 20% NHS, 20%decomplemented NHS, or GVBS++ alone in a volume of 1ml. Each mixture was chromatographed on Sephadex G-25,and 1-ml fractions were counted in a gamma counter to assay

TABLE 1. Bactericidal activity of NHS against six strains ofserum or resistant clinical isolates of E. coli and derived or

resistant serum clones

Initial Count/mlStrain (serotype) bacterial after serum % Killed

count/ml incubation

39 (04:K-12)Parent strain (serum S) 920 72 92Clonal derivative (serum R) 960 810 16

109 (06)Parent strain (serum S) 1,210 120 91Clonal derivative (serum R) 1,080 920 15

227 (018ac:K1)Parent strain (serum S) 660 60 91Clonal derivative (serum R) 840 3,600 0

248 (06:K2)Parent strain (serum S) 1,110 110 90Clonal derivative (serum R) 1,260 1,060 16

337 (075)Parent strain (serum S) 980 240 76Clonal derivative (serum R) 1,040 2,510 0

348 (07)Parent strain (serum S) 780 56 93Clonal derivative (serum R) 810 660 19

VOL. 55, 1987

on July 12, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

322 PORAT ET AL.

liposome-bound 51Cr and free 51Cr released from the lipo-somes.

RESULTSSelection of serum R clones. Serum R clones were selected

from each of six serum strains of E. coli of various 0-antigenserotypes (strain, serotype: 39, 04:K-12; 109, 06; 227,018ac:K1; 248, 06:K2; 337, 075; and 348, 07) by multiplepassages in serum-containing BHI broth and agar platescreening. The R clones were serially passed in serum-freemedium to ensure the stability of resistance, and identity ofthe parent S strains and R clones was confirmed by biotypeand antibiotic susceptibility patterns and serotyping (CentersiUlmaseikilloci20waaniselNJ

patyIsmtheco

TABLE 2. Comparison of intensity of bands in thehigher-molecular-weight regions of the PAGE gel

LPS Region aa Region ba Region Ca

109S 580 (60) 390 (40) 2.2 (0.2)R 468 (39) 584 (49) 40 (12)

227S 273 (66) 143 (34)R 790 (72) 315 (28)a Numbers represent absolute density readings.b Numbers in parentheses represent percentage of total intensity of all the

bands in the higher-molecular-weight regions.

r*iiseas iinri,r t*iain*a,* .Ja.). i aDIc I i11ustrates tIJ density of bands in samples heated to 100'C for 10 min (lanesignitude of bZacterial killing Of the parent serum S and.gnitudeobatrakiigoth2, 4, 6, and 8) did not differ from that of unheated samplesrum R clone by NHS. NHIS produced 90%/ or greaterrum R clone by NHS. NUS produced 90% ¢or greater (lanes 1, 3, 5, and 7). The major differences occur in the[ling with five of the SiX parent strains, While 75O killilngcuingdwithfve,ofxthe six3pare strains,whicon ,les7 klng banding pattern at the top of the gel (longer-chain, highlycurred with the sixth(337)strain. In contrast, lessan 0-antigen-substituted, higher-molecular-weight compo-% of the inoculum of four R clones (39, 109, 248, and 348).

, , ,, . ~nents), where LPS from R GNB exhibited more denselyis killed by NHS, while the remaining two R clones (227 n wstaining bands. Two dense bands (c) at the top of the gel ofd 337) multiplied. Two pairs of strains (109 and 227) were LPS from R 109 are barely visible in LPS from 5 109 (laneslected for further evaluation of the basis of resistance to 1 and 2). Scanning densitometry, performed through theH1S.PAGE. PAGE patterns of LPS from the two serum S courtesy of Fred Wingerath, LKB Instruments, Inc.(Rockville, Md.), confirmed the greater intensity of bands inrent strains and their R clones are illustrated in Fig. 1. Apical ladder formation of subunit structure of LPS from_oothGNBis seen, with densely staining broad bands at 109 and 227 than in LPS from S parent strains. Table 2 listsiooth GNB iS seen, with densely staining broad bandsat. .-..thecumulative density of three distinct clusters of bands,

e bottom of the gel, representing core components, ac- th cuuatv destbottomofthgel,representing core co, a- labeled a, b, and c on the PAGE gel in Fig. 1, in these areas,

unting for at least 50%/ of the ILPS. The number and confirming the greater content of more highly 0-antigen-substituted, high-molecular-weight subunits in LPS from Rclones than in LPS from their S parents. The c area (highest

1 2 3 4 5 8 7 8 molecular weight) density of R 109 exceeds that of S 109 byapproximately 20-fold.

c 1 p Lysis of liposomes containing LPS. (i) Liposome model. The

% of 51Cr ENTRAPMENTb b~~ ~_~~~~- -6i. ... 8i

S109 R109 S227 R227FIG. 1. PAGE of untreated LPS (lanes 1, 3, 5, and 7) and

samples heated to 1000C for 10 min (lanes 2, 4, 6, and 8) of twoserum S clinical isolates and R clones derived from them. Majordifferences are seen in the upper portion of the gels, areas a, b, andc from E. coli 109 and a and b from E. coli 227, representing thehigher-molecular-weight subunits of LPS. Much denser staining isseen in region a and b of the R clones than with serum S 109 and 227,while area c, the heaviest band of R 109, is barely seen in LPS fromits parent S strain.

eJ7SO xto R~~~~~~~~~~~~~IELEASE750

OL//.~~~~~~~~~~~~~

30 40 50 6o% of 51Qr RELEASE

FIG. 2. Dose-response curve of entrapment and release of 5"Crfrom liposomes prepared with various amounts of LPS after treat-ment with NHS. Percent entrapment and release of 5"Cr increasewith increasing doses of LPS with both approaching maximum atconcentrations slightly greater than 1 mg.

INFECT. IMMUN.

on July 12, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

LPS SUBUNITS AS DETERMINANTS OF SERUM RESISTANCE

I I I I30 40 50 60 70 80 90

Tube Number

I100

Tube Number

C.

FRESH NORMAL SERA

DECOMPLEMNTED SERA

2000r d.

IsooF-

1 soor

wI-

;

z

w

oI-

0U

I0

14001

1200

10001

80.0

eoo

4001

200

30 40 s0 60 70 s0 90 1UU

Tube Number

- .FRESH NORMAL *EAR

......uDECOMPLEMENTED OGRA

I __ I I I I I I10 20 30 40 s0 60 70 s0

Tube Number

FIG. 3. Distribution of 51Cr ,fter incubation of liposomes containing 51Cr and LPS. (a, S 109; b, S 227; c, R 109; d, R 227) with serum (solidline, NHS; broken line, heat-inactivated serum). Only a single peak, representing unlysed liposomes, is seen with heat-inactivated serum (a-d)with each LPS and with liposomes containing LPS from R 109 (c) and R 227 (d) incubated with NHS. Two peaks are seen with liposomescontaining LPS from S 109 (a) and S 227 (b) after incubation with NHS, with the second peak representing free 51Cr released from lysedliposomes.

oor a.

800s-

7001--

ziUlw

z

0U

LUcoz

0

- =FRESH NORMAL EPA.----- OECOMPLEAgNTED SERA

60o-

5001-

4001-

w-

z

iwa.co

I-0

to

300L

200t

500(

450(

400(

UlI~-

ziccwILCo

r-

z0

U.6S'

350(

300(

250(

200(

160C

100C

Soo

VOL. 55, 1987 323

.ft

on July 12, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

324 PORAT ET AL.

dose response of entrapment and release (Fig. 2) was deter-mined by measuring 5'Cr release after incubation of lipo-somes with variable amounts of S 227 LPS. Since bothentrapment and release approached maximum at an LPSconcentration of 1 mg, this dose was chosen for incorpora-tion for further studies. The applicability of these findings toLPS from R GNB was examined with 3H-radiolabeled LPSfrom S and R 227. Liposomes were prepared with 1-mgamounts of each [3H]LPS and separated from unincorpor-ated LPS by passage through a Sephacryl 1000 column (1.6by 40 cm; Pharmacia Fine Chemicals). Radioactivity, mea-sured in a liquid scintillation counter (Tracor Analytic Delta300) with Aquasol as the scintillator, demonstrated thatincorporation of [3H]LPS from both S and R 227 wasvirtually identical (12%).

(ii) Lytic assays. The differences in subunit composition ofLPS from serum S and serum R clones suggested, but did notestablish, LPS as the site of killing by NHS. The actual roleof LPS as the site of the bactericidal activity was establishedby using liposomes into which LPS from serum R and SGNB was incorporated along with 51Cr as a release markerfor lysis. Liposomes were prepared from LPS from twoserum S parent strains (S 109 and S 227) and two serum Rclones (R 109 and R 227) and separated from free 51Cr bypassage through a Sephadex G-25 column (1.6 by 30 cm).Separation curves demonstrated two peaks, with the firstpeak containing radiolabeled liposomes and the second peakcontaining free 51Cr. The liposome preparations werewashed with GVBS to remove excess LPS and suspended in

GVBS++, and 0.2-ml amounts were incubated in NHS,decomplemented NHS, or GVBS++ for 30 min at 35°C.Liposomes without LPS were not damaged and 51Cr was notreleased after incubation with NHS. Fig. 3a, b, c, and ddemonstrate results observed with the four LPS prepara-tions tested. (GVBS++ controls did not differ from resultswith decomplemented serum and are not shown.) Fig. 3a andb show release of 51Cr from liposomes containing LPS fromthe two serum S parent strains, with the initial peak repre-senting 5'Cr remaining in unlysed liposomes, while thesecond peak demonstrates release of approximately 50% ofincorporated 51Cr from lysed liposomes. In contrast, lipo-somes containing LPS from serum R clones (Fig. 3c and d)were not lysed by NHS, and no secondary peaks of 51Cr areseen. In all cases, incubation with decomplemented NHS didnot result in lysis of the liposomes and release of free51Cr.

Sephadex G-200 column chromatography of LPS. Sepha-dex G-200 chromatography of DOC-disaggregated LPS fromS and R GNB was performed, and each column fraction wasassayed for hexose (anthrone method) (Fig. 4, heavy solidlines) and KDO (Fig. 4, vertical lines), which was deter-mined qualitatively (- to + + + + +) since DOC precipitatedduring the reaction complicated quantitative determination(Fig. 4). Hexose content of the higher-molecular-weight Flfractions exceeded that of the F2 fractions from both strains,and the hexose content of the Fl fraction of LPS from Rclones greatly exceeded that of the Fl fraction from S parentstrains. The lower-molecular-weight fractions of LPS fromboth S and R GNB contain more KDO than the higher-molecular-weight fractions. The eluted fractions were

pooled into two large fractions, Fl (higher molecular weight)and F2, extensively dialyzed to remove DOC, and lyophi-lized. Table 3 lists the amounts recovered, the absolutequantities of hexose and KDO, and hexose/KDO ratios of allfour LPS preparations and their fractions. LPS from Rstrains contained approximately 2.5 times the amount of the

Fl fraction but less than 0.5 the quantity of the F2 fraction asLPS from S strains. Hexose/KDO ratios did not differmaterially between whole LPS of all strains or between theFl or F2 fractions from serum S or R LPS, respectively.However, hexose/KDO ratios of Fl fractions, approxi-mately 20:1, were four to six times greater than those of F2fractions, approximately 4:1. PAGE of each of these frac-tions demonstrated that the Fl and F2 fractions corre-sponded to the higher- and lower-moleQular weight regions(Fig. 5) previously described in Fig. 1.

Lysis of liposomes containing LPS fractions. These resultsled to direct examination of the role of the higher- andlower-molecular-weight LPS fractions in determining sus-ceptibility to lysis by NHS. Fractions Fl and F2 wereobtained from LPS from S and R strains, and 1 mg of eachwas incorporated into liposomes, as with the parent LPS,and treated with NHS. Similar concentrations of the Flfractions from both S and R E. coli 109 protected liposomesfrom lysis by NHS (Fig. 6a and b), with no release of 51Crfrom liposomes containing fraction Fl from LPS of eitherserum S or R strains. In contrast, liposomes containingfraction F2 from serum S and R strain 109 were lysed, and51Cr was released (Fig. 6c and d). Similar results wereobtained with Fl and F2 fractions from LPS of serum Sstrain 227 and its R clone. Liposomes containing Fl fractionsfrom both LPS preparations were not lysed, while thosecontaining the F2 fractions from both strains were lysed. Theamount of 51Cr released from liposomes containing F2 frac-tions were +60% which was slightly greater than that ofunfractionated LPS from serum S strains. Taken in toto,these observations provide convincing evidence that serumresistance in these isolates is mediated by the highly 0-antigen-substituted, higher-molecular-weight subunits ofLPS and that these are present in greater amounts in serumR GNB than in serum S GNB.

DISCUSSION

Since the description of Neisser and Wechsberg (15) of thebactericidal activity of NHS, almost all gram-negative bac-terial species have been shown to be killed by sera frommany animal species (26). Both the complement system andantibody appear necessary for killing of smooth strains of

TABLE 3. Hexose and KDO analysis of LPSs and their DOCcolumn fractions

Dried Hexosea KDOb RatioSample wt (mg) (,uglmg) (,ug/mg) Hexose/KDO

S 109 160 16 10.1Fl 14 230 11 21:1F2 36 170 36 5:1

R 109 180 23 8:1Fl 32 278 14 20:1F2 17 115 29 4:1

S 227 123 24 5:1Fl 13 190 10 18:1F2 37 70 21 3:1

R227 170 23 7:1Fl 34 210 10 20:1F2 16 125 39 3:1

a Cysteine-sulfuric acid method of Wright and Rebers (30).b Thiobarbituric acid method of Osbom (18).

INFECT. IMMUN.

on July 12, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

LPS SUBUNITS AS DETERMINANTS OF SERUM RESISTANCE 325

MDNC4J(O

0000

0-

CDGo

l<

Serum Resistant LPS Serum Sensitive LPS

Tube Number (2.5ml Fractions)FIG. 4. Fractionation of DOC-disaggregated LPS of serum R and S E. coli 227 on Sephadex G-200 and chemical analyses of the

higher-molecular-weight (F1) and lower-molecular-weight (F2) fractions for hexose (solid continuous line) and KDO (vertical bars). Corecomponents of LPS from both strains are not shown. OD 625, Optical density at 625 nm.

members of the family Enterobacteriaceae, although roughbacilli may be killed by complement components alone (6,15, 20, 27). The classical and alternative complement sys-tems both participate in serum killing, although the classicalpathway induces more rapid bactericidal activity than thealternative pathway (21).

Resistance to NHS is often considered a virulence factorsince GNB in the gastrointestinal tract are usually serum Swhile isolates from clinical infection are usually R (12, 19).Mechanisms responsible for the preponderance of serumresistance among clinical isolates have not been delineated,but these studies demonstrate that serum R clones can beselected from a serum S majority population by the killing ofsusceptible organisms by serum. Selection of phenotypicallystable R populations required serial passage but may occurmore rapidly in vivo than under the relatively artificialconditions of these studies. Rapid inactivation of comple-ment components in the medium used at the incubationtemperatures required might result in less complete elimina-tion of the S population than in clinical conditions in whichthere is continuous exposure to serum complement.

Several investigators have hypothesized that differences

.6

: :*

... FA

*:.. *

_ * ... : ^_ ;... _ ,

_ .__ -__

F1 F2 RLPS

FIG. 5. PAGE of LPS and Sephadex G-200 Fl and F2 fractionsfrom serum S and R E. coli 227 demonstrating the effectiveness ofseparation in that the Fl fraction of both strains was made up almostexclusively of higher-molecular-weight subunits, while the F2 frac-tion was exclusively lower-molecular-weight subunits. Subunitsrepresenting core components are partially excluded.

U _

SLPS

VOL. 55, 1987

I&PAW,CL, "

" .1 .,I

Fl F2

on July 12, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

326 PORAT ET AL.

4sor a.

600r b.- Fresh Normal Sore

.D,complomontedSere 700

0;eooEO~600

0 600

CLcx

, 40000

t-

O 300iso -

100_

Go0_

10 20 30 40 so 6o 70 60

Tube Number

2001

10(

~- NormaP@^@l@sera

* . .

.~~~~~~~~~~~~~~~~~~~~~~~1

10 20 30 40 50 60 70 60

Tube Number

- Fresh Normal Sore

.DecomplemontodSere

-

cx

C

CL0

00

la

s0or d.

7001-

- Fresh Normal Sera

.D*complementedSora

I

0oo0

5001

4001

3001

2001

10C

10 20 30 40 00 oU TU

Tube Number

FIG. 6. 51Cr release from liposomes prepared with Fl or F2 fractions of LPS from serum S and R E. coli 227 after incubation with NHS.No 51Cr release occurred after incubation with heated serum. Liposomes containing equivalent amounts of the Fl fraction of LPS from bothS (a) and R (b) E. coli 227 were not lysed, and no free 51Cr was released. In contrast, liposomes containing similar amounts of the F2 fractionfrom S 227 (c) and R 227 (d) were readily lysed by NHS as evidenced by the second peak of free 51Cr.

between serum S and R bacilli reside in the length of0-antigen chains of LPS (5, 22, 28). Many studies havecompared serum R strains with rough mutants (4, 22, 25, 28),whereas these studies compare serum S E. coli clinicalisolates with derived R clones. These clones were identicalto their parents serologically, biochemically, and by antibi-otic sensitivity pattern, but LPS from the R clones demon-

strated a greater number of longer-chain LPS molecules thanLPS of the parent S strain by PAGE.

Overall, hexose/KDO ratios, a measure of the relativeamount of core to 0-antigenic components, did not differsubstantially between LPS from S parent strains and their Rclones. However, scanning densitometry demonstrated thatLPS from serum R clones contained a higher proportion of

IF

400

0S

c

E 300

260CLc

U

,. 2000

V-

1**

soor C.

700oF-0

2 Goo

600CL0 GooO-c

012 400

e01-

.

1-.

200?-

1001

Tube Number

INFECT. IMMUN.

- - -I

OF

OFOF

on July 12, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

LPS SUBUNITS AS DETERMINANTS OF SERUM RESISTANCE 327

long-chain molecules than LPS from their S parents. Frac-tionation of disaggregated LPS confirmed that LPS from Rclones contained more long-chain subunits than LPS fromserum S strains. The longer, more highly 0-antigen-substituted LPS subunits contained the same proportion ofhexose to KDO (20:1) whether from serum S or R GNB.These differences in relative amounts of subunits of variouslengths between LPS from serum S and R GNB do notestablish that they, in themselves, are responsible for thevariations in serum sensitivity observed. Their significanceas determinants of sensitivity or resistance was established,however, by the use of liposomal membranes, simulatingbacterial cell walls, which incorporated LPS and 51Cr as arelease marker. LPS labeled with 3H from S and R GNBincorporated equally well into liposomes. Entrapment andrelease studies indicated that larger amounts of LPS wereneeded for optimal liposome formation than were actuallyincorporated, with 1 mg of LPS required to form 0.2 ml ofliposome suspension while only 12% was incorporated.Liposomes free of excess LPS and incubated with freshNHS were lysed, releasing approximately 50% of incorpo-rated 51Cr when S LPS was used. Liposomes with no LPS orcontaining LPS from serum R clones were not lysed byserum, and serum heated to 56°C had no effect on any of theliposomes. Even more striking results were obtained withliposomes incorporating long- or short-chain G-200 fractionsof LPS. Liposomes containing equivalent concentrations ofthe Fl or long-chain LPS subunit fractions from LPS ofeither S or R GNB behaved like those containing LPS fromR GNB in their resistance to lysis by serum. In contrast,liposomes containing F2 or short-chain fractions of LPSfrom either S or R GNB were readily lysed by serum, similarto liposomes containing LPS from serum S GNB. Thesefindings firmly establish LPS as one reactive site for killing ofGNB by NHS. They also demonstrate that resistance tokilling by NHS is mediated by the content of longer-chain,high 0-substituted LPS subunits. The exact mechanism bywhich these longer-chain subunits block bactericidal activityhas yet to be established but may result from the combina-tion of antibody and complement components at a site toodistant from the cell surface to allow the membrane attackcomplex to damage the cell membrane. Whether this iscorrect or whether complement activity is blocked in an-other manner has yet to be determined, but this liposomalmodel of bacterial cell membranes provides a model thatallows more specific delineation of the role of individualbacterial components and their subunits in the serum bacte-ricidal reaction.

ACKNOWLEDGMENTS

We thank Raquel Rodriguez for her assistance in the preparationof the manuscript.

This study was supported in part by Public Health Service grantAl 21783 from the National Institute of Allergy and InfectiousDiseases.

LITERATURE CITED1. Craven, D. E., S. Bruins, and W. R. McCabe. 1977. Sepsis due

to Gram-negative bacilli: epidemiology, pathogenesis, and im-munologic aspects. In I. M. Ledingham (ed.), Recent advancesin intensive therapy. Churchill Livingstone, Ltd., Edinburgh.

2. Dupont, H. L., S. B. Formal, R. B. Hornick, M. J. Snyder, J. P.Libonati, D. G. Sheahan, E. H. Labrec, and J. P. Kalas. 1971.Pathogenesis of Escherichia coli diarrhea. N. Engl. J. Med.285:1-9.

3. Fierer, J., F. Finley, and A. I. Braude. 1972. A plaque assay onagar for detection of Gram-negative bacilli sensitive to comple-

ment. J. Immunol. 109:1156-1158.4. Goldman, J. N., S. Ruddy, K. F. Austen, and D. S. Feingold.

1969. The serum bactericidal reaction. III. Antibody and com-plement requirements for killing a rough Escherichia coli. J.Immunol. 102:1379-1387.

5. Goldman, R. C., K. Joiner, and L. Leive. 1984. Serum-resistantmutants of Escherichia coli 0111 contain increased lipopolysac-charide, lack an 0-antigen-containing capsule, and cover moreof their lipid A core with 0-antigen. J. Bacteriol. 159:877-882.

6. Inoue, K., K. Yonemasu, A. Takamizawa, and T. Amano. 1968.Studies on the immune bacteriolysis. XIV. Requirement of allnine components of complement for immune bacteriolysis.Biken J. 11:203-206.

7. Johanson, W. G., Jr., D. E. Woode, and T. R. Chaudhuri. 1979.Association of respiratory tract colonization with adherence ofGram-negative bacilli to epithelial cells. J. Infect. Dis.139:667-673.

8. Kataoka, T., K. Inoue, 0. Luderitz, and S. C. Kinsky. 1971.Antibody and complement-dependent damage to liposomes pre-pared with bacterial lipopolysaccharides. Eur. J. Biochem.21:80-85.

9. Kinsky, S. C., J. A. Haxby, D. A. Zopf, C. R. Alving, and C. B.Kinsky. 1969. Complement-dependent damage to liposomesprepared from pure lipids and Forssman hapten. Biochemistry8:4149-4158.

10. Laemmli, U. K. 1970. Cleavage of structural proteins during theassembly of the head of bacteriophage T4. Nature (London)227:680-685.

11. McCabe, W. R. 1973. Gram-negative bacteremia. Disease-a-Month. Yearbook Medical Publishers, Chicago.

12. McCabe, W. R., B. Kajser, S. Olling, M. Uwaydeh, and L. A.Hanson. 1978. E. coli in bacteremia: K and 0 antigens andserum sensitivity of strains from adults and neonates. J. Infect.Dis. 138:33-41.

13. Mintz, C. S., M. A. Apicella, and S. A. Morse. 1984. Electro-phoresis and serological characterization of the lipopolysaccha-ride produced by Neisseria gonorrhoeae. J. Infect. Dis.149:544-552.

14. Moll, A., P. A. Manning, and K. N. Timmis. 1980. Plasmid-determined resistance to serum bactericidal activity: a majorouter membrane protein, the tra T gene product, is responsiblefor plasmid-specified serum resistance in Escherichia coli. In-fect. Immun. 28:359-367.

15. Neisser, M., and F. Wechsberg. 1901. Uber die Wirkungartbactericider sera. Muench. Med. Wochenschr. 48:697-700.

16. Ogata, R. T., and R. P. Levine. 1980. Characterization ofcomplement resistance in Escherichia coli conferred by theantibiotic resistance plasmid R 100. J. Immunol. 125:1494-1498.

17. Opal, S., A. Cross, and P. Gemski. 1982. K antigen and serumsensitivity of rough Escherichia coli. Infect. Immun. 37:956-960.

18. Osborn, M. J. 1963. Studies on the gram-negative cell wall. I.Evidence for the role of 2-keto-3-deoxyoctonate in the lipopoly-saccharide of Salmonella typhimurium. Proc. Natl. Acad. Sci.USA 50:499-506.

19. Roantree, R. J., and L. A. Rantz. 1960. A study of the relation-ship of the normal bactericidal activity of human serum tobacterial activity. J. Clin. Invest. 39:72-81.

20. Robbins, J. B., K. Kenny, and E. Suter. 1965. The isolation andbiological activities of rabbit yM- and yG- anti-Salmonellatyphimurium antibodies. J. Exp. Med. 122:385-402.

21. Root, R. K., L. Eliman, and M. M. Frank. 1972. Bactericidaland opsonic properties of C4 deficient guinea pig serum. J.Immunol. 109:477-486.

22. Rowley, D. 1968. Sensitivity of rough gram-negative bacteria tothe bactericidal action of serum. J. Bacteriol. 95:1647-1650.

23. Scott, T. A., and E. H. Melvin. 1953. Determination of dextranwith anthrone. Anal. Chem. 25:1656-1661.

24. Svanborg-Eden, C., E. C. Gotschlich, and T. K. Korhonen. 1983.Aspects of structure and function of pili of uropathogenic E.coli. Prog. Allergy 33:189-200.

25. Taylor, P. W. 1975. Genetical studies of serum resistance inEscherichia coli. J. Gen. Microbiol. 89:57-66.

VOL. 55, 1987

on July 12, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

INFECT. IMMUN.

26. Taylor, P. W. 1983. Bactericidal and bacteriolytic activity ofserum against gram-negtive bacteria. Microbiol. Rev. 47:46-83.

27. Vukajlovich, S. W., and D. C. Morrison. 1983. Conversion oflipopolysaccharides to molecular aggregates with reducedsubunit heterogeneity: demonstration of LPS-responsiveness in"endotoxin-unresponsive" C3H/HeJ splenocytes. J. Immunol.130:2804-2808.

28. Wardlaw, A. C. 1963. The complement-dependent bacteriolytic

activity of normal human serum. II. Cell wall composition ofsensitive and resistant strains. Can. J. Microbiol. 159:41-52.

29. Westphal, O., 0. Luderitz, and F. Bister. 1952. Uber dieExtraktion von Bacterien mit Phenol/Wasser. Z. Naturforsch.76:148-152.

30. Wright, B. G., and P. A. Rebers. 1972. Procedure for determin-ing heptose and hexose in lipopolysaccharides. Modification ofthe cysteine-sulfuric acid method. Anal. Biochem. 49:307-319.

328 PORAT ET AL.

on July 12, 2020 by guesthttp://iai.asm

.org/D

ownloaded from